TLS and DNS in practice: what a public audit of 51 state portals shows
Every state government portal answers on 443. Only two thirds negotiate TLS 1.3, and a quarter validate DNSSEC. Here is what that means and how to check your own domains.
TLS and DNS are the two layers everyone assumes are handled. We checked 51 state government portals to see what handled actually looks like. The answer is better than a decade ago and worse than the assumption.
- 51 of 51 answer on port 443.
- 50 of 51 present a chain that validates.
- 34 of 51 negotiate TLS 1.3 (66.7%).
- 13 of 51 validate DNSSEC (25.5%).
- 9 of 51 publish AAAA records (17.6%).
- 4 of 51 have a certificate expiring within 30 days of the scan.
TLS 1.2 is not a failure, but it is a floor
TLS 1.2 with a modern cipher suite is still sound. The reason to care about 1.3 is not that 1.2 is broken, it is that 1.3 removes the negotiation surface where downgrade and legacy-cipher problems live, and it cuts a round trip out of the handshake.
A third of the cohort still stops at 1.2. That is usually a load balancer or appliance that has not been updated, not a deliberate choice. The fix is a configuration change, and the only way to know you need it is to ask what the server actually negotiates rather than what the config file says.
One useful negative result: no portal in the cohort accepted TLS 1.1. The 2018 deprecation has largely landed on public-sector infrastructure, which is worth noting because it is usually cited as the thing that never gets done.
Certificate expiry is the boring outage
Four portals had a certificate expiring within 30 days of the scan. Short-lived certificates and automated renewal have made this better, but the failure mode has not gone away: an expiry is still one of the most common ways a public site goes down, and it is entirely preventable with a calendar entry or an automated check.
The lesson is not to trust renewal automation blindly. Check the expiry date from outside, where the user sits, not from the server that holds the key.
DNSSEC is the real laggard
Three quarters of the cohort does not validate DNSSEC. This is the weakest number in the dataset by a wide margin, and it is where the gap between configured and protected is largest.
DNSSEC signs DNS answers so a resolver can tell a forged response from a real one. Without it, cache poisoning and DNS hijacking do not require breaking any encryption, only intercepting a query. The operational cost is real, particularly around key rollover, which is why adoption stalls. That does not change the exposure.
How to check your own domains
- Ask the server what it negotiates:
curl -sI --tlsv1.3 https://your.domain. If it fails, try--tlsv1.2and compare. - Read the chain and expiry:
openssl s_client -connect your.domain:443 -servername your.domain. The output includes the chain and the validity dates. - Check DNSSEC with
dig +dnssec your.domain. A validated answer carries theadflag when the resolver validates; an unsigned zone returns no RRSIG at all. - Check for AAAA records with
dig AAAA your.domainif IPv6 matters for your users.
Our TLS certificate checker handles the first two from a browser and reports the negotiated version, chain validity, and days to expiry. The DNS lookup tool covers the CNAME, MX, and TXT records that usually explain why something else is broken.
Worked example: one host, end to end
Here is a real handshake against a host that gets this right, with every line that matters:
$ echo | openssl s_client -connect example.com:443 -servername example.com -brief
CONNECTION ESTABLISHED
Protocol version: TLSv1.3
Ciphersuite: TLS_AES_256_GCM_SHA384
Peer certificate: CN = example.com
Hash used: SHA256
Signature type: ECDSA
Verification: OK
Server Temp Key: X25519, 253 bits
Four lines carry the diagnosis. Protocol version says 1.3, so there is no downgrade surface to worry about. Peer certificate: CN = example.com has to match the name you asked for, or the browser will warn even though the chain is valid. Verification: OK means the chain built and the roots trusted it. Server Temp Key tells you the key exchange actually negotiated rather than one the server claims to support.
Compare with a weaker host, where the same command reports Protocol version: TLSv1.2 and a certificate whose CN is a different name than the one you connected to. Both of those produce a working connection in a browser and a broken one from a script that validates properly, which is why checking from the server you actually run from matters.
For DNSSEC, the flag to look for is ad:
$ dig +dnssec example.com
;; flags: qr rd ra; QUERY: 1, ANSWER: 2, AUTHORITY: 0, ADDITIONAL: 1
No ad here. That does not prove the zone is unsigned, and it does not prove it is signed either. It means the resolver you asked did not return a validated answer, which could be an unsigned zone or a resolver with validation disabled. To separate those, query a resolver you know validates, or read the zone's DS record at the parent. Treating a missing ad as "DNSSEC is broken" is the most common misreading of this command.
The full dataset, including per-state rows, is on the state government security study page. Method: public DNS records plus one TLS handshake per host, run on 5 October 2026 against the 51 state and DC government primary portals. No port scanning, no vulnerability probing, no crawling. Figures describe that snapshot and nothing later.
Keep reading
- Email authentication in practice: what 51 state governments get wrong92% of state government portals publish a DMARC record. Only 29% enforce it. Here is what that gap means, how SPF, DKIM and DMARC fit together, and how to check a domain in about a minute.
- Reading the results: IPs, subnets, ports, and WHOISWhat each lookup actually tells you, and how to interpret the answer. A practical reference for IP addresses, subnet math, open ports, and registration records.